n_distinct(result$MA.id) # number of MAs
## [1] 23492
nrow(result) # 9 model result per each MA
## [1] 211428
table((result %>% distinct(MA.id, .keep_all = TRUE))$event)
## 
##     common infrequent       rare  very rare 
##      22074       1154        101        163
proportions(table((result %>% distinct(MA.id, .keep_all = TRUE))$event)) * 100
## 
##     common infrequent       rare  very rare 
## 93.9639026  4.9123106  0.4299336  0.6938532

estimation

Analysis of heterogeneity estimate

There are three types of model that have \(\tau\)

  1. \(\tau\) from ivre, htevaguelogit and htelogit_tau50 (log ratios)

  1. \(\tau_a\) and \(\tau_c\) from ablogit and ablogit_tau50 (log odds)

  1. \(\tau_a\) and \(\tau_c\) from htebeta (prob)

Correlation of tau

cor(
  result_bayesian[result_bayesian$modelname == "htelogit_tau50",]$tau,
  result_bayesian[result_bayesian$modelname == "ablogit_tau50",]$taua
  )
## [1] 0.2477361
cor(
  result_bayesian[result_bayesian$modelname == "htelogit_tau50",]$tau,
  result_bayesian[result_bayesian$modelname == "ablogit_tau50",]$tauc
  )
## [1] 0.3046824

cor(
  result_bayesian[result_bayesian$modelname == "htebeta",]$taua,
  result_bayesian[result_bayesian$modelname == "ablogit_tau50",]$taua
  )
## [1] -0.08585503
cor(
  result_bayesian[result_bayesian$modelname == "htebeta",]$tauc,
  result_bayesian[result_bayesian$modelname == "ablogit_tau50",]$tauc
  )
## [1] -0.1115232

Analysis Significance Inference by model

modelname Sig_sum percentage_total
ivfe 10403 44.283160
ivre 9132 38.872808
ctevaguelogit 11553 49.178444
ctebeta 11402 48.535672
htevaguelogit 7308 31.108462
htelogit_tau50 6047 25.740678
htebeta 4363 18.572280
ablogit 8003 34.066916
ablogit_tau50 597 2.541291 

table(result_sigsum$Sig_sum)
## 
##     0     1     2     3     4     5     6     7     8     9 
## 10947  1075   706  1407  1587   922   915  2312  3121   500
round(proportions(table(result_sigsum$Sig_sum))*100,3)
## 
##      0      1      2      3      4      5      6      7      8      9 
## 46.599  4.576  3.005  5.989  6.755  3.925  3.895  9.842 13.285  2.128
table(result_sigsum$event, result_sigsum$Sig_sum)
##             
##                 0    1    2    3    4    5    6    7    8    9
##   common     9948  977  632 1340 1524  902  883 2261 3109  498
##   infrequent  799   74   58   59   61   18   29   42   12    2
##   rare         65    8    7    5    2    2    3    9    0    0
##   very rare   135   16    9    3    0    0    0    0    0    0
round(proportions(table(result_sigsum$event, result_sigsum$Sig_sum), margin = 1)*100,2)
##             
##                  0     1     2     3     4     5     6     7     8     9
##   common     45.07  4.43  2.86  6.07  6.90  4.09  4.00 10.24 14.08  2.26
##   infrequent 69.24  6.41  5.03  5.11  5.29  1.56  2.51  3.64  1.04  0.17
##   rare       64.36  7.92  6.93  4.95  1.98  1.98  2.97  8.91  0.00  0.00
##   very rare  82.82  9.82  5.52  1.84  0.00  0.00  0.00  0.00  0.00  0.00

waic

For best model comparison among seven Bayesian models (exclude ivfe and ivre)